The Use of Multiple Methods to Explore Three-Way Person, Brand and Usage Context Interactions

Niraj Dawar, INSEAD
S. Ratneshwar, University of Florida
Alan G. Sawyer, University of Florida
ABSTRACT - Previous research on situational influences focused on two-way product-usage context interactions and generally employed single methods. In contrast, the present research investigated three-way interactions between person, product, and usage context, and used multiple methods. Specifically, we studied such interactions in the product category bar soaps, with gender serving as the relevant person-level segmentation variable. The results of three experiments provided converging evidence that males, in comparison to females, are less discriminating between different subcategories of soap, and are more likely to cross-use brands in those subcategories across different usage contexts.
[ to cite ]:
Niraj Dawar, S. Ratneshwar, and Alan G. Sawyer (1992) ,"The Use of Multiple Methods to Explore Three-Way Person, Brand and Usage Context Interactions", in NA - Advances in Consumer Research Volume 19, eds. John F. Sherry, Jr. and Brian Sternthal, Provo, UT : Association for Consumer Research, Pages: 116-122.

Advances in Consumer Research Volume 19, 1992      Pages 116-122

THE USE OF MULTIPLE METHODS TO EXPLORE THREE-WAY PERSON, BRAND AND USAGE CONTEXT INTERACTIONS

Niraj Dawar, INSEAD

S. Ratneshwar, University of Florida

Alan G. Sawyer, University of Florida

ABSTRACT -

Previous research on situational influences focused on two-way product-usage context interactions and generally employed single methods. In contrast, the present research investigated three-way interactions between person, product, and usage context, and used multiple methods. Specifically, we studied such interactions in the product category bar soaps, with gender serving as the relevant person-level segmentation variable. The results of three experiments provided converging evidence that males, in comparison to females, are less discriminating between different subcategories of soap, and are more likely to cross-use brands in those subcategories across different usage contexts.

INTRODUCTION

Consumers' choices of brand or product often depends on the situation or usage context for which the product or brand is desired. This finding has been established repeatedly in a number of studies that have emphasized the importance of product-situation interactions in explaining consumer choice and in determining market structures (e.g., Belk 1974, 1975; Dickson 1982; Miller and Ginter 1979; Srivastava, Alpert and Shocker 1984). It has also been argued (e.g., Dickson 1982) that in some cases it may be useful to examine person-situation-product interactions, where some product (or brand) choices (or attitudes) are influenced by the usage context and person variables differentially than other products (or brands). For example, a brand of soap may be designated by female users as a "shower" soap and another may be designated a "facial" soap, while male users may not make the distinction.

The person-situation-product interaction can be important since it has the potential for becoming a basis for segmentation if it can be shown that the groups identified by using this criterion have homogeneous within-group responses and heterogeneous between group responses (Dickson 1982). In the soap-brand example above, a person-situation-product interaction might identify two segments of consumers based on a person variable (gender) whose behavior (product choice) varies differentially across different situations or usage contexts (shower vs. facial usage). The behavior within each group or segment (females or males) is homogeneous in that the product x situation interaction can explain it.

While such three-way person x product x usage context interactions were conceptualized by previous researchers (e.g., Dickson 1982; Srivastava et al. 1984), to the best of our knowledge they have never been demonstrated empirically. Research in this area has invariably treated all subjects as replicate experimental units; in other words, subjects were not "blocked" on the basis of relevant person-level variables. (This is not surprising, given that much of this research focused on the product x usage context interaction). In this present research we sought to investigate such three-way interactions in the product category bar soaps, with gender serving as the relevant person-level variable. Based on preliminary research, we expected to find systematic differences between male and female consumers in their category structures; that is, in their mental representations of the product category. More importantly, we also expected that these gender differences in category structures would be closely associated with differences in perceptions of appropriateness of various brands of soap for different usage situations (Ratneshwar & Shocker 1991).

The present research had one additional unique feature, namely, the use of multiple methods. Previous researchers of usage situation influences on consumers' product preferences invariably adopted a single method: typically, one wherein all subjects rated a researcher-specified list of product alternatives across a set of usage contexts. There is some possibility that this method induces demand biases (Reingen 1976; Sawyer 1975). Specifically, Reingen (1976) suggested that the design of these studies (e.g. Belk 1974) may have cued subjects that they were "supposed to" shift their preferences across usage contexts. Hence, we decided on the use of multiple methods to examine whether converging evidence could be obtained on the three-way person-product-situation interaction.

We report three experiments. In Experiment 1, we explored whether males and females differ in their mental representations of various soap brands with the data being obtained from a categorization task. In a second experiment we investigated gender differences in terms of soap brands evoked from memory in different usage contexts. Finally, in Experiment 3 these gender differences were examined in a task where subjects rated a specified list of brands for appropriateness in different usages.

EXPERIMENT 1

A focus group (N=12) was first conducted to obtain some initial insights into gender differences in the manner in which consumers think about the product category. The discussion in the group suggested that in general males were not likely to be as sensitive as females to the usage situation for bar soaps. Concommitantly, males (vs. females) did not appear to discriminate as much between different subtypes of soaps. These conjectures led us to believe that males might differ from females in their mental representations of the product category, and that these differences might be associated with their respective perceptions of usage appropriateness of different brands. These speculations are consistent with recent research in categorization that suggests that category structures are not fixed across populations (Barsalou 1987), and that such cognitive structures can be goal-derived (Barsalou 1985; and see Ratneshwar and Shocker 1991). This implies that different subpopulations may use different categorization schemes to classify the same group of objects. For example, males and females may differ in the manner of classifying brands of soap because female consumers are more sensitive to usage goals, and are perhaps more expert in the soap category. As a result, they may use finer classifications than do males; that is, they may have more differentiated category structures. Also, the distinctions that they make among various subcategories of soap are likely to be clearer in that, as experts, they are likely to "avoid confusions between brands..." (Alba & Hutchinson 1987). In addition, as experts, their "basic" level might be more specific than that of novices and consequently, their categorical distinctions might be at a lower level than that of novices (Alba & Hutchinson 1987). Because of such differences in categorization, we expected that males might have a broader classification (fewer subcategories, with a larger number of brands of soap per subcategory) than females. Further, we expected that distinctions between brands belonging to different subcategories of soap might be coarser (i.e. less in magnitude) for males than females. Experiment 1 was designed to explore these issues.

FIGURE

GENDER DIFFERENCES IN PERCEIVED DISSIMILARITY BETWEEN SOAP SUBCATEGORIES

Method

Twenty one brands of bar soap that were judged to be familiar by the focus group participants were utilized for the experiment (see Table 3 for the list of brands). A group of 32 female and 30 male undergraduate subjects participated. Subjects were asked to sort 21 cards, each carrying the name of a brand, into no more than eight groups based on perceived similarity. These data were converted into interbrand dissimilarity measures; these dissimilarity measures were computed at the aggregate level separately for each gender (see Ratneshwar and Shocker, 1991, Study 2, for details).

Results

The dissimilarity measures from the male and female groups were used as input to hierarchical cluster analyses and there were no major discrepancies between males and females as to the brands in each cluster. The results of this analysis indicated that both genders clustered the brands into three basic clusters. Based on packaging information, the brands in these three clusters closely corresponded to the usage subcategories Facial, Deodorant, and General Purpose soaps. By analyzing the differences between males and females on perceived dissimilarity of brands, we could determine if males did indeed perceive the brands to be less dissimilar than did females. The dissimilarity of brands that fell into different subcategories formed the data of interest. We analyzed these data to check for effects of Gender in perceived dissimilarity between soap subcategories.

The mean dissimilarity measures for both genders are presented in the Figure. Males and females alike perceived a large difference between Facial brands and Deodorant brands, although females perceived the difference to be greater (t58=5.98, p<.0001). With regard to the other comparisons, the General Purpose to Deodorant and General Purpose to Facial, the data show that males perceived greater similarity across categories than did females (t118=8.5, p<.0001; and t98=14.64, p<.0001 respectively). Thus, in all cases, females perceived greater dissimilarity across subcategories than did males.

Discussion

The results supported our expectation that males would differ from females in their perceptions of similarity of various subcategories of soap brands. Further, the extent of these differences apparently depends on which two subcategories are compared. Since males tend to perceive all brands of soap as more similar than females, they may be more likely to cross-use soaps than females. For example, males may be more likely to use a General Purpose soap for facial usage than females. Further, it is more likely that males may cross-use soaps across the Facial and General Purpose subcategories than across the Facial and Deodorant subcategories. The results of this experiment are consistent with the argument that as experts, females' distinctions among the subcategories of soap are likely to be clearer, and they are less likely to confuse brands. The category structures with which brands of soap are represented in memory apparently are different for males and females. This suggests that the principal cognitive tasks associated with such memory structures, such as brands evoked from memory, are also likely to be affected by gender differences. This issue was investigated in Experiment 2.

EXPERIMENT 2

This experiment examined potential three-way person x product x usage context interactions in the formation of evoked sets. As numerous studies demonstrate, the formation of an evoked set or consideration set, as some authors prefer to call it, is a necessary stage in the phased decision process (Alba & Chattopadhyay 1985, Bettman 1979, Howard & Sheth 1969, Ratneshwar & Shocker 1991, Wright & Barbour 1977). Most previous research into the effects of usage-context on consumer behavior has not considered the evoked-set formation stage, as subjects in these studies were provided with lists of product alternatives (see Ratneshwar and Shocker, 1991, Study 3 for an exception).

Evoked sets are subsets of a larger pool of brands or alternatives represented in category structures in memory (Alba & Chattopadhyay 1985, Bettman 1979). The products that make up these category structures vary in how typical they are of that category, and typicality is a major determinant of recall of category members from memory (Medin & Smith 1984). Recent research has found that typicality may vary as a function of context and of subject population (Barsalou 1985, 1987, Roth & Shoben 1983). That is, categories are now thought to be flexible enough to allow for different instances to represent those categories depending, among other things, on the context in which they are evoked, and on the population evoking them (Dawar 1991). Evoked set formation is equivalent to the retrieval of instances of a subcategory from memory in a particular usage context (Ratneshwar & Shocker 1991). We can, therefore, expect that evoked sets may vary across subjects based on some subject variable as well as across usage context. In this experiment we examined how evoked sets differ across subject populations and usage contexts. Specifically, we investigated gender differences in the types of soap brands evoked by males and females in two different usage contexts (facial usage and shower usage).

Method

Subjects were two different groups of undergraduate students. The first group was made up of 68 marketing majors while the second was composed of 81 psychology majors. Subjects were later blocked on the basis of gender. All subjects evoked soap brands in both usage contexts, making usage context a within-subjects factor. Subjects were asked to name "brands of soap that you might use regularly to wash your face (on your body while taking a shower)". Order of presentation of usage context was counterbalanced across subjects. Subjects were asked to limit the number of brands listed to a maximum of three in each usage context. This was done so as to obtain only those brands that subjects were most likely to retrieve from memory in that usage context.

Subjects' brand listings were classified by the researchers into three types of soap subcategories based on Experiment 1: (1) General purpose, (2) Facial, and (3) Deodorant. A cross tabulation of the brand recall frequencies into the 24 cells generated by crossing Usage Context (two levels) with Groups nested in Gender (four levels) and Soap Subcategory (three levels) was analyzed using Analysis of Variance (ANOVA).

Results

The mean frequencies with which brands were evoked in the different subcategories in each usage context are shown in Table 1. An ANOVA was conducted using "group" as the experimental block. Since there are replicate groups within each gender, we can test all main effects and interactions including the three-way interaction of person-product-usage context. The three-way interaction of gender, usage context and brand recall was marginally significant (F2,4=5.8, p<.07). Some clear trends are apparent in terms of gender differences. For facial usage, females (vs. males) are much more likely to use facial soaps, and much less likely to use deodorant soaps. For shower usage, females (vs males) are much more likely to use general purpose soaps, and much less likely, again, to use deodorant soaps. Other interactions were significant too. The significant interaction of Product x Usage Context (F2,4=42.85, p<.01) confirms previous findings (Ratneshwar and Shocker 1991) that brand recall is dependent on usage context, this time in the category of bar soaps. The significant interaction of the Gender and Product variables (F2,4=10.33, p<.05) shows that brand recall in this product category also depends on the gender of the consumer -- males and females differ in their recall of different types of soap. The lack of significance of the Usage Context x Gender interaction (F1,2=.02, p>.75) indicates that the number of brands of soap recalled does not vary by gender with usage context. Further, the lack of significance of the main effect for Gender (p>.50) shows that variance due to gender may, in large measure, be accounted for by the Gender x Product and Gender x Product x Usage Context interactions.

TABLE 1

EXPERIMENT 2: MEAN FREQUENCIES WITH WHICH BRANDS IN DIFFERENT SUBCATEGORIES WERE EVOKED IN DIFFERENT USAGE CONTEXTS

Discussion

The results provided a confirmation of previous research by demonstrating the importance of the two-way usage context x brand interactions. However, the marginally significant three-way interaction is probably the most crucial aspect of the present results. Prior studies have generally not looked at the manner in which usage context differences may occur at the stage of evoked-set formation. The results obtained here suggest that differences between segments of consumers may arise at the evoked-set formation stage of the choice process. This implies that later stages (evaluation, choice) may all be affected by the interactions at the evoked-set stage. In Experiment 3 we examine brand appropriateness ratings for a three-way person x product x usage context interaction. This is the method traditionally adopted in studies of situation effects on consumer behavior.

EXPERIMENT 3

The previous two experiments attempted to examine the three-way person x product x usage context interaction by examining differences in perceived dissimilarity and brands of soap evoked in various usage contexts. In this experiment we adopted a method that has traditionally been used in the study of situational effects on buyer behavior (Belk 1974, Srivastava et al. 1984). In this method subjects are asked to rate the appropriateness of a set of product alternatives for each of a number of usage situations. This yields a Product x Situations x Subjects experimental design. The data so obtained can be analyzed to estimate the contribution to variance of each effect and their interactions. But as noted previously, in past research all subjects have been treated as replicates and individual difference variables have been typically ignored. However, judging by the results of Experiments 1 and 2, it would appear that Gender is an important person-level variable on the basis of which one might expect a person x product x usage context interaction in the bar soap category. Note that the earlier two experiments dealt with earlier stages of the consumer choice process, namely, representation of alternatives and evoked set formation. This experiment is different from the previous two in that the dependent measure used here is much closer to actual behavior (Wicker 1972). In this study we examine the three-way interaction and its magnitude in consumers' ratings of appropriateness of brands of soap to each of three usage contexts.

Method

The 21 brands of soap chosen for the study were generated in a focus group and selected based on subjects' familiarity with them (see Experiment 1). The usage contexts used in this experiment were also generated in the focus group. The three used in this study were: (a) a brand of soap that you might use regularly to wash your face; (b) a brand of soap you might use regularly to wash your hands; (c) a brand of soap you might use regularly on your body while taking a shower. A total of 82 ( 48 male and 34 female) undergraduate students participated in the experiment for course credit. Subjects rated the appropriateness of each brand (on a seven-point scale) in each usage context, making both brands and situations within-subject factors. The order of presentation of the three usage contexts was counterbalanced across subjects.

TABLE 2

EXPERIMENT 3: CONTRIBUTIONS TO VARIANCE IN BRAND USAGE APPROPRIATENESS RATINGS

Results

The analysis aims to estimate the relative contribution to variance in appropriateness ratings of each of the main effects and interactions. This is accomplished by using the expected mean squares of each of the effects (see Belk 1974). The relative contributions of each of the effects and interactions are presented in Table 2. As anticipated, the three-way Gender x Brands x Usage Context interaction was significant (p<.001) with 2.9% contribution to overall variance. However the most important sources of variance are the Brands x Usage Context interaction (26.8%, p<.001), and the main effect for Brands (14.9%, p<.001). The contribution of the main effects of Gender and Usage Context are small. These results are consistent with earlier research on the topic of situational effects, even though the alternatives used in these earlier studies were products rather than brands, and the person variables used were different. Table 3 presents the means of the appropriateness ratings. These means show clearly that males and females tend to have a different pattern of response with regard to brand usage appropriateness ratings.

Stated simply, the data appear to suggest that females are more discriminating than males in their judgement of soaps, and that their judgements depend on the usage context. Males, generally, perceive fewer differences between soap subcategories and would be more prone to cross-use. For example, females rate deodorant brands as most appropriate for shower usage (mean=5.99), only moderately appropriate for use for washing hands (mean=5.19), and not very appropriate for facial use (mean=2.29). While males follow the same ordinal pattern in terms of appropriateness ratings, the mean ratings do not differ as much across usage context. For example, males rate deodorant soaps as reasonably appropriate for all three situations (mean=6.24 for shower, mean=5.81 for hands, and mean=4.08 for facial use).

GENERAL DISCUSSION

The three experiments reported here provided converging results pointing toward the importance of an interaction between person variables, usage context, and alternatives, in explaining consumer behavior. The three-way interaction of Gender, Usage Context, and Brands was marginally statistically significant in Experiment 2 and statistically significant in Experiment 3. An examination of the cell means for Experiment 3 shows that males tend to rate deodorant brands as being more appropriate in all usage contexts than do females. This is also reflected in the result from Experiment 2 in that males tend to recall more deodorant brands across usage contexts than do females. The results from both experiments seem to show that males' preferences for deodorant brands show very little variations across contexts, while females tend to judge their brands with regard to specific usage context. It is further apparent from the results of both Experiments 2 and 3, that females have a stronger preference than males to use general purpose brands for showering. These findings are corroborated by the results of Experiment 1, which investigated the perceived similarity of brands across subcategories. Females tended to discriminate more between brands based on the typical usage of those brands, while males tend to view these subcategories as more similar.

Unlike previous research in this area ( e.g. Belk 1974, Srivastava et al. 1984), we employed multiple methods to assess the impact of three-way person-usage context-brand interactions. In the first Experiment, we examined differences in perceived similarity of brands across gender, using a card-sort task to generate dissimilarity measures. The second experiment used a method that asked subjects to generate alternatives from memory for different usage contexts. Finally, in Experiment 3, we reverted to the traditional method, wherein subjects rated a set of brands across different usage contexts. Thus, each of the three experiments in this study examined the three-way person x product x usage context interaction at a different stage of the consumer choice process. Our results show that the three-way interaction can explain differences in the mental representation of soap brands for males vs. females (Experiment 1); differences in the evoked set of soap brands considered in different usage contexts (Experiment 2) and differences in the evaluation or appropriateness judgements (Experiment 3). Previous research has generally only examined the last of these stages of the consumer choice process.

TABLE 3

EXPERIMENT 3: CELL MEANS OF BRAND USAGE APPROPRIATENESS RATINGS

In addition, our operationalizations of the person, product, and situation variables differed somewhat from those used in previous research. The person variable used in this study was gender, the product variable was brands of soap, and the usage context was three usage situations for soap. In spite of these differences, the results of Experiment 2 and especially Experiment 3 are remarkably consistent with the results of earlier studies. Specifically, the importance of the two-way interaction of Brands x Usage Context, and Brands x Person, is evident in this study as it has been in previous studies. Further, the converging pattern of results -- notwithstanding the differences in method -- from three experiments suggests that such interactions are "real," and not due to methodological artifact (cf. Reingen 1976).

Previous research has pointed out the importance of usage context interaction in determining customer segments and defining market structures (Belk 1974; Dickson 1982; Ratneshwar & Shocker 1991). The three-way person x product x usage context interaction has the potential for becoming a basis for segmentation if it can be shown that the groups identified through this criterion have homogeneous within-group responses and heterogeneous responses between groups (Dickson 1982). The results in this study indicate that this can indeed be the case. As the results show, the person by situation segments that could be identified on the basis of these results are more homogeneous within segment than across segment. Specifically, these results suggest that the mental representations, evoked set formation, and appropriateness judgements are specific to gender x situation groups. It is useful for marketers to understand that some segments (e.g. women) may be more sensitive to usage context differences than other segments (e.g. men) for some products. In addition, some brands (e.g. deodorant brands) may be more influenced by person x situation factors than others (e.g. facial and general purpose brands).

Categories and category structures, especially in marketing situations are often goal-derived. Usage contexts may serve to "frame" or define consumer goals by specifying the specific benefits sought. Thus, perceptions of usage appropriateness may depend on the extent to which females or males perceive that a particular brand will provide the specific benefits sought in that context. These usage appropriateness perceptions, in turn are likely to impact on the mental representation of categories.

Finally, we note that the present research attests to the importance of research on the interaction between person, usage context, and product variables at various stages of the consumer problem-solving process. Future research may be able to verify whether there are systematic differences in the kinds of interactions that explain each stage of the problem solving task, across a person variable such as experts vs. novices. For example, experts may take into consideration a greater number of factors than novices in making their evaluations, but may be just as heuristic as novices in the creation of evoked sets.

REFERENCES

Alba, Joseph W., and Amitava Chattopadhyay (1985), "Effects of Context and Category Cues on Recall of Competing Brands," Journal of Marketing Research, 22 (August), 340-349.

Alba, Joseph W. and J. Wesley Hutchinson (1987), "Dimensions of Consumer Expertise," Journal of Consumer Research, 13 (March), 411-454.

Barsalou, Lawrence W. (1985), "Ideals, Central Tendency, and Frequency of Instantiation as Determinants of Graded Structure in Categories," Journal of Experimental Psychology - Learning, Memory and Cognition, 11 (4), 629-654.

Barsalou, Lawrence W. (1987), "The Instability of Graded Structure: Implications for the Nature of Concepts," in Concepts and Conceptual Development: Ecological and Intellectual Factors in Categorization, Ulric Neisser ed., Cambridge: Cambridge University Press, 101-140.

Belk, Russell W. (1974), "An Exploratory Assessment of Situational Effects in Buyer Behavior," Journal of Marketing Research, 11 (May), 156-163.

Belk, Russell W. (1975), "Situational Variables and Consumer Behavior," Journal of Consumer Research, 2 (December), 157-164.

Bettman, James R. (1979), An Information Processing Theory of Consumer Choice, Reading, MA:Addison-Wesley.

Dickson, Peter R. (1982), "Person-Situation: Segmentation's Missing Link," Journal of Marketing, 46 (Fall), 56-64.

Howard, John A., and Jagdish N. Sheth (1969), The Theory of Buyer Behavior, New York: Wiley.

Loken, Barbara and James Ward (1990), "Alternative Approaches to Understanding the Determinants of Typicality," Journal of Consumer Research, 17 (2), 111-126.

Miller, Keith E. and James L. Ginter (1979), "An Investigation of Situational Variation on Brand Choice Behavior and Attitude," Journal of Marketing Research, 16, (February), 111-123.

Ratneshwar, S. and Allan D. Shocker (1991), "Substitution in Use and The Role of Usage Context in Product Category Structures," Journal of Marketing Research, 28 (August), 281-295.

Reingen, Peter H. (1976), "Demand Bias in the Assessment of Situational Effects on Buyer Behavior," in Advances in Consumer Research, Vol. 3 (Ann Arbor, MI: Association for Consumer Research).

Rosenberg, Seymour, Carnot Nelson and P. S. Vivekananthan (1968), "A Multidimensional Approach to the Structure of Personality Impressions," Journal of Personality and Social Psychology, 9 (4), 283-294.

Roth, Edward M. and Shoben E. J. (1983), "The Effect of Context on the Structure of Categories," Cognitive Psychology, 16, 371-416.

Sawyer, Alan G. (1975), "Demand Artifacts in Laboratory Experiments in Consumer Research," Journal of Consumer Research, 1 (March), 20-30.

Srivastava, Rajendra K, Mark I. Alpert, and Allan D. Shocker (1984), "A Customer-oriented Approach for Determining Market Structures," Journal of Marketing, 48 (Spring), 32-45.

Wicker, Allen W. (1972), "Processes Which Mediate Behavior-Environment Congruence," Behavioral Science, 17 (May), 265-276.

Wright, Peter, and Frederick Barbour (1977), "Phased Decision Strategies: Sequels to Initial Screening," in M. Starr and M. Zelny (eds), Multiple Criteria Decision Making, North Holland TIMS Studies in the Management Science. Amsterdam: North Holland, 91-109.

----------------------------------------